Glottochronology

node

The mid-twentieth-century attempt, associated above all with Morris Swadesh, to date language splits by assuming that basic vocabulary is replaced at a roughly constant rate. Take a fixed list of core meanings, count how many of them are cognate between two languages, and convert the retention rate into elapsed time with a universal constant calibrated on languages with written histories. The appeal was obvious — it promised absolute dates where the comparative method gives only relative ones — and the method was rapidly and comprehensively criticised. Replacement rates are not constant across meanings, languages or periods; cultural upheaval and contact accelerate them; the constant was calibrated on a handful of Indo-European cases and generalised to the world; and cognacy judgements, which the arithmetic treats as data, are themselves the contested output of prior comparative work. By the 1970s glottochronology was largely abandoned in mainstream historical linguistics. Its underlying question did not go away, and the computational phylogenetic methods that returned in the 2000s address the same problem with explicit models of rate variation — which is why their critics so often reach first for the word glottochronology.

A node is not a place. Drawing it on a map would assert something about the world that no stored fact supports.